That doesn't seem at all straight-forward to me. The whole point of a classical liberal education (which is the kind of education that most college students are paying for) is give a student a broad background in addition to an area of specialization. I majored in mathematics as an undergrad, but I wouldn't consider my geology, anthropology, and music classes to be "fluff", and I don't think that an art history major should consider their mathematics classes to be "filler" (where both "fluff" and "filler" seem to imply that the classes are not important).

So, as I suggested above, tracking a student's major GPA vs the general GPA makes sense, but declaring that all non-major classes are unimportant filler and shouldn't matter at all goes a bit too far---if I am looking at candidates for a job opening or trying to decide who is going to get into a graduate program, if I had two candidates that were the same on paper, except one did better in their non-major classes, I would consider that student the better candidate. Those are data that I would want to have.

I assign to a student the grade which that student has earned. Can you suggest a better verb for the action that I am performing?

This is incorrect. From the horse's mouth:

This means that when a student of any age enrolls in a postsecondary institution, such as a community college or university, the rights of parents to access educational records passes to the student, and the parents no longer have rights to such access. It doesn't matter who is footing the bill, or if the student is a dependent major.

Even if they aren't an adult, I can't share any information. FERPA covers the educational privacy of minors, too.

A U shape is unrealistic. This assumes that the only difference between students is their level of dedication and how well they pay attention. The reality is that not all students are created equal, and some are stronger than others. Indeed, there is some reason to believe that students might be normally distributed. Yes, we expect college students to come from the elite of society, but some will be more elite than others, and since not everyone is majoring in the same field, we would expect the lower division classes (at least) to include both strong students and weak students. In any case, remember that you are not comparing university students to a "general large population," but to the general population of university students.

Personally, when I am assigning grades, I expect there to be very few As and Fs (one should have to work very hard to get an A, and one has to be particularly negligent to earn an F), a fair number of Bs and Ds, and a large number of Cs (where a C means that a student has demonstrated adequate performance, but nothing noteworthy). I don't force the grades to fall into a normal distribution, but they do normally fit something that looks approximately normal (though a beta distribution might be a better model, given the fact that grades fall on a fixed interval---and the distribution is rarely unimodal, meaning that neither a beta nor normal distribution is a very good model). That said, I have had classes where more than half of the students get a B or better, and I had one very depressing semester where less than half of the class managed to get a C (huzzah for lower division math classes that are required for graduation with any major).

This. A thousand times, this. This is the reason that I assign grades based on where the scores cluster, rather than on a completely arbitrary number (though I do use those arbitrary numbers as a guide).

It should be noted that your scheme really is very arbitrary. I'm glad it worked out for you, but mixing medians and standard deviations simply don't make any kind of statistical sense. The IQR (inter-quartile range) would probably be a better measure of spread if you are going to use the median as a measure of center. One should also note that your scheme is biased in favor of As over Fs. Perhaps that is what you intend, though I personally prefer that the median correspond to the center of the C range, rather than the top.

One of the nice thing about teaching at a college or university is that the faculty don't have to deal with helicopter parents. Parents can call all they want, and all they should ever hears is "I'm sorry, but it would be a violation of students' FERPA rights for me to divulge any information to you." Pissy students are another matter entirely.

My institution does not seem to have access to a digital version of the above linked paper. If anyone else has institutional access and can get a digital copy, I, too, would like to see it.

Which classes are the filler and fluff, and which classes are real classes? Who makes that decision? A better solution might be to track a person's major GPA separately from their total combined GPA (many graduate schools ask for this, anyway).

Grading on a curve is grading on a percentile rank. The curve merely defines which letter corresponds to which rank.

Um... the OP was suggesting that the marketing campaign was original, not the book.

And here I thought they were a bit more expensive than a grand or two.

That isn't what the post to which you are replying says at all. He didn't marry just any random person who would be willing to kill him. He (presumably) married someone with whom he was/is compatible on any number of issues (desire for children, sexual preferences, social status, religion, cultural background, and so on). Of these many points of comparison, a common desire for a particular kind of end-of-life care was a deal breaker; i.e. even if the OP's spouse had been compatible on many other issues, the willingness to pull the plug (or more actively kill) was important. It is no different than someone choosing not to marry someone because they want children and their partner does not; or because they are Catholic and their partner is Jewish.

A: To hold the foreskin back.